Deep Learning Approach to Automatic Detection of Tool Wear in Machining Using Coolants

The project advances the development of an AI-based system for automatic cutting tool condition characterization using machine learning and machine vision. It addresses the limitations of indirect tool wear monitoring by enabling direct, image-based wear analysis both on- and off-machine, with a focus on reducing downtime and improving tool utilization. A key challenge is adapting the system to industrial environments with cooling lubricants, which requires changing camera hardware and compensating for reduced image quality through advanced image enhancement techniques. Building on a previous model that for wear segmentation and classification under dry machining conditions, the project evaluates model weighting strategies and extends the approach to lubricated processes. The dataset and predictive maintenance models are expanded and re-evaluated to improve tool life prediction beyond simple linear regression. Overall, the project aims to deliver robust data pipelines, model architectures, and preprocessing methods that support accurate wear progression analysis and practical Industry 4.0 deployment

Faculty Supervisor:

Dan Wu

Student:

Partner:

Karlsruher Institut für Technologie

Discipline:

Computer science

Sector:

Education

University:

University of Windsor

Program: